Upload
shannon-mclaughlin
View
213
Download
0
Embed Size (px)
Citation preview
Application of Emotion Recognition Methodsin Automotive Research
Matthias Wimmer, Christian Peter, Jörg Voskamp, Martin A. Tischler
Institut für Informatik
2nd Workshop “Emotion and Computing”, 10.9.2007, Osnabrück
Tischler, Peter, Wimmer & Voskamp, Application of Emotion Recognition Methods in Automotive Research, 2nd Workshop "Emotion and Computing"
2
Theoretical Background of Driving Pleasure
Driver‘s Objective: Optimal Activation
Task Demands
Driveability
Sportiveness
Comfort
Capability
Task DifficultyResources und
Restrictions
Veh
icle
Environment
Emotional estimation
Transfer ofDriver‘s Request
Jordan (2000): “Four Pleasures” Physio-pleasure Psycho-pleasure
Socio-pleasure Ideo-pleasure
based onFuller (2005)
Tischler, Peter, Wimmer & Voskamp, Application of Emotion Recognition Methods in Automotive Research, 2nd Workshop "Emotion and Computing"
3
Measurement of Emotions in Vehicles
EMOTION
Subjectivecomponent
Expressivecomponent
Behavioralcomponent
Verbal report Driving dynamics
Physiologicalcomponent
Cognitivecomponent
Video/Audio-recording
Measurement Verbal report
Facial expressions
Gestures
Changes in voice
ECG, EMG, EDAQuestioning during driving
Driving style and operator control action
Cognitive appraisal of the vehicle and the situation
Tischler, Peter, Wimmer & Voskamp, Application of Emotion Recognition Methods in Automotive Research, 2nd Workshop "Emotion and Computing"
4
Pilot Study „Driving Pleasure W204“
2007 C220 CDI (W204) 1982 190E (W201)
Tischler, Peter, Wimmer & Voskamp, Application of Emotion Recognition Methods in Automotive Research, 2nd Workshop "Emotion and Computing"
5
Cooperation
Tischler, Peter, Wimmer & Voskamp, Application of Emotion Recognition Methods in Automotive Research, 2nd Workshop "Emotion and Computing"
6
Subjects: 8 non professional drivers (age: 33-53)
Tischler, Peter, Wimmer & Voskamp, Application of Emotion Recognition Methods in Automotive Research, 2nd Workshop "Emotion and Computing"
7
Task:Drive on three courses
Bosch Proofing Ground Boxberg(near Heilbronn)
Autobahn
Country road
Handling course
Tischler, Peter, Wimmer & Voskamp, Application of Emotion Recognition Methods in Automotive Research, 2nd Workshop "Emotion and Computing"
8
Technical Setup
driving dynamics
meterEREC glove
video camera
microphone
cell phonehandsfree set EREC
recording unit
Tischler, Peter, Wimmer & Voskamp, Application of Emotion Recognition Methods in Automotive Research, 2nd Workshop "Emotion and Computing"
9
The Methods in Detail
Physiological measurement system EREC (skin resistance, heart rate, temperature)
Interviews and questionnaires after
driving each car
Video and audio recorder on the
back seat
Drivers were asked to “think aloud”
Tischler, Peter, Wimmer & Voskamp, Application of Emotion Recognition Methods in Automotive Research, 2nd Workshop "Emotion and Computing"
10
Subjective Driving Experience
1,0 3,0 5,0 7,0
subjective control
no worries about safety
drove too risky
vehicle is predictable
vehicle met my desires
MB 190EMB C220
strongly agreestrongly disagree
N=8, Method: questionnaire
Tischler, Peter, Wimmer & Voskamp, Application of Emotion Recognition Methods in Automotive Research, 2nd Workshop "Emotion and Computing"
11
Facial Expression RecognitionInstitut für Informatik
Mercedes-Benz 190EMean expression of happiness: 4,2% Peak value: 15,0%
Mercedes-Benz C-Class 220Mean expression of happiness: 5,5% Peak value: 19,8%
0% = neutral face100% = maximum expression
Tischler, Peter, Wimmer & Voskamp, Application of Emotion Recognition Methods in Automotive Research, 2nd Workshop "Emotion and Computing"
12
1
1 3 5 7 9
A
C
H
A H C
MB 190E
MB C-Class 220
A AutobahnC Country roadH Handling
Valence
Activation
Speech Analysis
+
+
-
-
Most important parameters• changes of pitch• intensity• energy changes over frequency bands
Average confidence1.1-1.2 for valence0.9-1.0 for arousal
Tischler, Peter, Wimmer & Voskamp, Application of Emotion Recognition Methods in Automotive Research, 2nd Workshop "Emotion and Computing"
13
Conclusions
Methods show corresponding results
Speech analysis
Easy to apply, but speech is not a natural product while driving
Facial expression recognition
Challenge in fitting the face model and classifying the facial expression
Problems: changing light and background, non emotional-related head movements, driver is talking
Advantages of tested methods
Non disturbing
Continuous data collection
Useful completion to psychological methods